Quantification of free coumarin and its liberation from glucosylated precursors by stable isotope dilution assays based on liquid chromatography-tandem mass spectrometry detection

Article abstract of DOI:10.1021/jf0728348

A stable isotope dilution assay for the quantification of free coumarin and glucosylated coumarin precursors has been developed using [¹³C ₂]-coumarin as the internal standard. The doubly labeled coumarin was synthesized by reacting [¹³C₂]-acetic anhydride with salicylic aldehyde and characterized by means of mass spectrometry and nuclear magnetic resonance (NMR) experiments. The specifity of liquid chromatography-tandem mass spectrometry enabled unequivocal determination and sensitive quantitation of the odorant. Because of the very simple extraction procedure, free coumarin could be analyzed within 1h. For quantification of total coumarin, the odorant was liberated from its precursors by an incubation with hydrochloric acid or β-glucosidase. In analyses of breakfast cereals, the intra-assay coefficient of variation was 9.9% (n = 5) for total coumarin. When coumarin was added to butter cookies at a level of 10 μg/kg, a recovery of 94.1% was found. Further addition studies revealed a detection limit of 2.9 μg/kg and a quantification limit of 8.6 μg/kg. Application of the stable isotope dilution assay to several plants, foods, and essential oils revealed high contents in cassia products and those foods in which cassia has been used as an ingredient. In contrast to this, Ceylon cinnamon contained much less coumarin. The odorant was also quantified in woodruff, clover seeds, and the essential oils of lavender, citron, and chamomile. Only trace amounts were detected in carrots and the essential oils of peppermint and dill, whereas in bilberries, black raspberries, and Angelica roots, coumarin was below detectable levels. In Ceylon cinnamon and cassia, the odorant occurred mainly in its free form, whereas in fenugreek seeds and woodruff, 68 and 88% of the total coumarin content was liberated from glucosylated precursors, respectively.

Full text of DOI:10.1021/jf0728348

796 J. Agric. Food Chem. 2008, 56, 796–801
Quantification of Free Coumarin and Its Liberation
from Glucosylated Precursors by Stable Isotope
Dilution Assays Based on Liquid
Chromatography-Tandem Mass Spectrometric
Detection
MICHAEL RYCHLIK*
Lehrstuhl für Lebensmittelchemie der Technischen Universität München, Lichtenbergstrasse 4,
D-85748 Garching, Germany
A stable isotope dilution assay for the quantification of free coumarin and glucosylated coumarin
precursors has been developed using [13C2]-coumarin as the internal standard. The doubly labeled
coumarin was synthesized by reacting [13C2]-acetic anhydride with salicylic aldehyde and characterized
by means of mass spectrometry and nuclear magnetic resonance (NMR) experiments. The specifity
of liquid chromatography-tandem mass spectrometry enabled unequivocal determination and sensitive
quantitation of the odorant. Because of the very simple extraction procedure, free coumarin could be
analyzed within 1h. For quantification of total coumarin, the odorant was liberated from its precursors
by an incubation with hydrochloric acid or ꢀ-glucosidase. In analyses of breakfast cereals, the intra-
assay coefficient of variation was 9.9% (n ) 5) for total coumarin. When coumarin was added to
butter cookies at a level of 10 µg/kg, a recovery of 94.1% was found. Further addition studies revealed
a detection limit of 2.9 µg/kg and a quantification limit of 8.6 µg/kg. Application of the stable isotope
dilution assay to several plants, foods, and essential oils revealed high contents in cassia products
and those foods in which cassia has been used as an ingredient. In contrast to this, Ceylon cinnamon
contained much less coumarin. The odorant was also quantified in woodruff, clover seeds, and the
essential oils of lavender, citron, and chamomile. Only trace amounts were detected in carrots and
the essential oils of peppermint and dill, whereas in bilberries, black raspberries, and Angelica roots,
coumarin was below detectable levels. In Ceylon cinnamon and cassia, the odorant occurred mainly
in its free form, whereas in fenugreek seeds and woodruff, 68 and 88% of the total coumarin content
was liberated from glucosylated precursors, respectively.
KEYWORDS: Cassia; cinnamon; coumarin; glucosides; liquid chromatography-tandem mass spectro-
metry; stable isotope dilution assay
INTRODUCTION
which was suspected to form DNA adducts and may react to
hepatotoxic o-hydroxyphenylacetaldehyde (3). Whereas the
former reaction was found not to be the cause for carcinogenic
effects in rodents, the latter product was confirmed to evoke
hepatotoxicity by coumarin (4). For these reasons, a maximum
level of 2 mg/kg for foods generally and 10 mg/kg in alcoholic
beverages has been set in the European Union (5). Moreover,
coumarin is not allowed to be used as flavoring additive to foods.
The odorant coumarin is a natural component of several
spices, the respective essential oils, and other flavoring foods,
such as Cinnamomum aromaticum (cassia bark), Asperula
odorata (sweet woodruff), Dipterix odorata (tonka bean), and
species of clover. Besides occurring naturally in these foods,
coumarin has been widely used as a flavoring compound because
of its sweet and aromatic odor. However, since the early 1950s,
the odorant has been found to exert hepatotoxicity and was
suspected to be mutagenic and carcinogenic (1). Although
coumarin in humans is mainly metabolized to 7-hydroxycou-
marin (2), a subpopulation lacks this detoxification pathway and,
by contrast, metabolizes the odorant to its 3,4-epoxy derivative,
In the last decades, cassia bark increasingly substituted true
cinnamon in baked goods, particularly in seasonal products, such
as gingerbread or cinnamon star cookies. Moreover, indications
to use cassia bark powder as a supplement and remedy against
type 2 diabetes mellitus (6) increased consumption of this spice
in the Western countries. Because a tolerable daily intake of
0.1 mg/kg body weight has been established by the Scientific
Panel on Food Additives, Flavourings, Processing Aids, and
* To whom correspondence should be addressed. Telephone: +49-
89-289-132-55. Fax: +49-89-289-141-83. E-mail: michael.rychlik@
ch.tum.de.
10.1021/jf0728348 CCC: $40.75
2008 American Chemical Society
Published on Web 01/16/2008

Stable Isotope Dilution Assay of Coumarin
J. Agric. Food Chem., Vol. 56, No. 3, 2008 797
Materials in Contact with Food (AFC) (7), a longer lasting
consumption of products high in cassia can be expected to
provoke hepatotoxic effects. Therefore, supplements containing
cassia have been classified as drugs, and coumarin provisionally
had been restricted to 67 mg/kg in cinnamon star cookies and
50 mg/kg in gingerbread in Germany during the winter season
in 2006 until November first. Moreover, the consumption of
cinnamon star cookies by children had been recommended not
to exceed four cookies per day.
In view of these concerns, analytical methods are required
for accurate and sensitive quantitation of coumarin in baked
goods or spices.The most frequently used method for quantifying
coumarin is a high-performance liquid chromatography (HPLC)
assay with UV detection (8, 9). However, because the detection
limit of HPLC has been reported to be as high as approximately
2 mg/kg (10), the latter method appeared not sensitive enough
to verify the compliance of foods with the legal limit of 2 mg/
kg. Moreover, complex matrices, such as baked goods, require
more accurate methodologies for quantitation. Gas chromato-
graphy-mass spectrometry (GC-MS) methods are also known
(11) but have been little used until now because of the need for
extraction with organic solvents and for separating the odorant
from nonvolatile matrix compounds (12).
For a convenient cleanup and unambigueous detection of
many low-volatile compounds, LC coupled to mass detection
has gained increasing importance. However, it is generally
accepted that cleanup is likely to cause losses of the analyte
and ionization efficiency in liquid chromatography-mass
spectrometry (LC-MS) strongly depends upon coeluting matrix
compounds (13). Therefore, quantitation of food samples is more
accurate if an internal standard (IS) is used, which has very
similar chemical and physical properties and behaves nearly
identically throughout the whole analytical procedure. Therefore,
stable isotopologues of the analytes are considered the best IS
in LC-MS. Analoguously to our reports on the quantitation
of vitamins (14) and trichothecene mycotoxins (15), the use of
labeled analogues furthermore allows for the compensation of
losses and enables the most accurate quantitation.
Therefore, the aim of the present investigation was to
synthesize an isotopologue of coumarin and apply it as an
internal standard (IS) for quantitation of coumarin using
LC-MS detection. Furthermore, because coumarin is known
to occur in some plants mainly bound as a glucosylated
precursor, an additional aim of this study was to quantify bound
coumarin as well.
Figure 1. Mass spectrum of (A) coumarin and (B) [13C2]-coumarin in
electron impact ionization.
Germany) and a mixture of diethyl ether and pentane (1:1, v/v) as the
mobile phase. Three fractions at Rf of 0.45, 0.56, and 0.83 were visible
on the TLC plate, and the zone with Rf ) 0.45, representing [13C2]-
coumarin, was scratched from the plate. The resulting powder was
suspended in diethyl ether (2 mL), and the suspension was filtered
yielding the pure product. The two fractions at Rf 0.56 and 0.83 were
suspended in water (5 mL); the suspension was filtered and, after the
addition of sulfuric acid (60% w/w, 1 mL), the resulting solution was
heated for 5 h at 150 °C. Separation of [13C2]-coumarin was performed
as described above, and this procedure was repeated another 2 times
until no product was generated. Purity of the product (4.2 mg; 11.5%)
was checked by GC-MS and LC-MS/MS.
Mass spectra in electron impact ionization and positive electrospray
LC-MS/MS are shown in Figures 1B and 2B, respectively.
13C NMR (CDCl3) δ: 116.1 (d, 1JCC ) 70.2, C-3), 166.2 (d, 1JCC
)
71.2, C-2).
1
2
3
1H NMR (CDCl3) δ: 6.42 (ddd, JHC ) 170, JHC ) 8, JHH ) 10,
2
H-3), 7.33 (m, H-5-H-8), 7.60 (m, H-5-H-8), 7.93 (dd, JHC ) 7,
3JHH ) 10, H-4).
Stable Isotope Dilution Assay (SIDA) for the Determination of
Free Coumarin in Foods. Baked goods, spices, or herbs were minced
in a blender (Privileg, Quelle, Fürth). Samples (0.1 g) with considerable
amounts of glucosylated coumarin were homogenized in a mixture of
methanol/saturated CaCl2 (2 mL, 80:20, v/v) by means of an Ultraturrax.
The resulting powders (0.5-0.01 g) or homogenates (2 g) were stirred
for 1 h at 20 °C in aqueous methanol (80%, 5 mL) or a mixture of
methanol/saturated CaCl2 (80:20, v/v, 5 mL), respectively, containing
MATERIALS AND METHODS
Chemicals. The following chemicals were obtained commercially:
[
13C2]-acetic anhydride, chloroform, coumarin, and salicylic aldehyde
(Aldrich, Steinheim, Germany); acetonitrile, CaCl2, diethyl ether, formic
acid, hydrochloric acid, methanol, pentane, sodium sulfate, and sulfuric
acid (Merck, Darmstadt, Germany); and ꢀ-glucosidase from almonds
(Sigma, Deisenhofen, Germany).
[
13C2]-coumarin (20 ng-10 µg).
Synthesis of [13C2]-Coumarin. [13C]-Labeled coumarin was pre-
pared by a modification of the synthetic procedure to unlabeled
coumarin by Perkin (16). Salicylic aldehyde (30 mg, 246 µmol),
aqueous sulfuric acid (60% w/w, 1 mL), and [13C2]-acetic anhydride
(13 mg, 123 µmol) were mixed in a closable vial; the latter was purged
with nitrogen and heated for 6 h at 150 °C. Subsequently, water (2
mL) was added to the mixture, which was then transferred in a
separation funnel. The resulting solution was extracted with chloroform
(3 × 5 mL), and the organic phases were dried over anhydrous sodium
sulfate. After the solvent was evaporated, the residue was dissolved in
diethyl ether (1 mL) and purified by preparative thin-layer chroma-
tography (TLC) using silica gel with fluorescence detection as the
stationary phase (silica gel 60, 0.25 mm, F254, Merck, Darmstadt,
The extracts were filtered and, after passing through a 0.4 µm syringe
filter (Millipore, Bedford, MA), analyzed by LC-MS/MS.
Hydrolysis of Glucosides for Quantification of Total Coumarin.
For liberation of bound coumarin precursors, the homogenized samples
were stirred either in hydrochloric acid (2.5 mol/L, 5 mL) at 80 °C for
90 min or in a solution of ꢀ-glucosidase from almonds (1 mg/mL, 5
mL) at 37 °C for 60 min. Subsequently, the extracts were filtered and,
after passing through a 0.4 µm syringe filter (Millipore, Bedford, MA),
analyzed by LC-MS/MS.
Liquid Chromatography-Tandem Mass Spectrometry (LC-MS/
MS). LC-MS/MS was performed by means of a triple quadrupole
Finnigan TSQ Quantum Discovery (Thermo Electron Corporation,
Waltham, MA) coupled to a Finnigan Surveyor Plus HPLC System

798 J. Agric. Food Chem., Vol. 56, No. 3, 2008
Rychlik
Figure 3. Synthetic route to [13C2]-coumarin.
Determination of Detection and Quantification Limits. A total
of 2, 5, 10, and 20 ng of coumarin were added to ground butter cookies
(0.5 g) and analyzed as detailed before. Each sample was analyzed in
triplicate. Detection (DL) and quantification limits (QL) were calculated
according to Hädrich and Vogelgesang (18).
Precision and Recovery. Intra-assay precision was evaluated by
analyzing breakfast cereals in five extracts as detailed before.
Recovery was determined by adding 5 ng of coumarin to ground
butter cookies (0.5 g) and performing SIDA as detailed before in
quadruplicate analysis.
NMR. 13C NMR spectra were recorded with an AM 360 spectrom-
eter (Bruker, Karlsruhe, Germany): transmitter frequency, 90.56 MHz;
spectral width, 23809 Hz; repetition time, 2.5 s; 256 scans; 64 K data
set; 1-Hz line broadening. Processing was performed by multiplication
with a Lorentz–Gaussian function prior to transformation. Chemical
shifts are expressed in parts per million downfield from tetramethyl-
silane, and J values are in hertz.
Figure 2. LC-MS/MS spectrum of (A) coumarin and (B) [13C2]-coumarin
after CID of the protonated molecules in positive electrospray ionization
mode.
RESULTS AND DISCUSSION
Synthesis of [13C2]-Coumarin. A previous study on the
synthesis and application of [13C]-labeled coumarin has been
reported by Christakopoulos et al. (12). However, the latter
authors prepared only singly labeled coumarin, which suffers
from a spectral overlap of 10% in the molecular ions of labeled
and unlabeled coumarin because of natural isotopic abundance
in unlabeled coumarin. To overcome this lack and because of
the commercial availability of fully labeled [13C]-acetic anhy-
dride, we prepared doubly labeled coumarin according to a
modification of Perkin’s initial route to unlabeled coumarin (16)
(Figure 3). We obtained [13C2]-coumarin in 11.5% yield,
showing in electron impact ionization a negligible overlap in
its molecular ion signal at m/z 148 with unlabeled coumarin, as
low as 2%. The correct incorporation of the labels was verified
by 13C NMR spectrometry, which revealed two doublet signals
corresponding to the labeled positions C-2 and C-3 of coumarin
in comparison to the reference spectrum of unlabeled
coumarin.
Mass Spectrometry of Coumarin in Electron Impact
Ionization. In accordance with the two [13C]-labels introduced,
labeled coumarin showed a shift of the molecular ion signal to
m/z 148 compared to the respective signal at m/z 146 of
unlabeled coumarin (Figure 1). The two further intense signals
of [13C2]-coumarin at m/z 119 and 90 were consistent with the
loss of two molecules of 13CO from the lactone ring to form
the radical cation (3) (Figure 4A). In the case of unlabeled
coumarin, these losses result in signals at m/z 118 and 90, the
latter of which is identical with the respective ion of the labeled
compound. This pathway is in good accordance with the
assumptions of Christakopoulos et al. (12) and Porter (19).
Mass Spectrometry of Coumarin in Electrospray Ioniza-
tion. For the anticipated stable isotope dilution assay (SIDA)
with LC-MS detection, mass spectrometry after atmospheric
pressure ionization was applied for coumarin detection. Most
suitable was the positive electrospray mode, giving an intense
signal of the protonated molecule at m/z 149 and 147 for [13C2]-
coumarin and unlabeled coumarin, respectively.
(Thermo Electron Corporation, Waltham, MA) equipped with a 150
× 2 mm i.d., 5 µm, Aqua C-18 reversed-phase column (Phenomenex,
Aschaffenburg, Germany). A total of 10 µL of the sample solutions
were chromatographed using gradient elution with variable mixtures
of aqueous formic acid (0.1%, eluent A) and formic acid in acetonitrile
(0.1%, eluent B), at a flow of 0.2 mL/min. A 20 min linear gradient
was programmed from 0 to 100% B. Then, 100% B was maintained
for 3 min and subsequently brought back within 1 min to 0% B and
held for another 15 min to allow for column equilibration. During the
first 14 min of the gradient program, the column effluent was diverted
to waste to ensure an adequate spray stability. For [13C2]-coumarin,
the mass transition (m/z precursor ion/m/z product ion) 149/104 and,
for unlabeled coumarin, the mass transition 147/103 were chosen. Spray
voltage was set to 3500 V; sheath gas pressure was 35 mTorr; and
auxiliary gas pressure was 5 mTorr. Capillary temperature was 350
°C, and capillary offset was 35 V. Source collision-induced dissociation
(CID) was used with the collision energy set at 12 V. For LC-MS/
MS experiments, the collision gas pressure in quadrupole 2 was 1.5
mTorr; the scan time was 0.20 s; and the peak width in quadrupole 1
and 3 were adjusted to (0.7 amu. The collision energy for coumarin
isotopologues was set to 16 V.
GC-MS Analysis. Mass chromatograms in the electron impact (EI)
mode were recorded by means of a MD 800 quadrupole mass
spectrometer (Fisons Instruments, Manchester, U.K.) coupled to a type
8000 gas chromatograph (ThermoQuest, Egelsbach, Germany) equipped
with a 30 m × 0.32 mm i.d., 0.25 µm, fused silica capillary DB-5
(Fisons Instruments, Mainz, Germany). The samples were applied by
split injection at 230 °C and a split ratio of 1:20. After the sample (2
µL) was injected, the temperature of the oven was raised from 40 to
250 °C at a rate of 10 °C/min and held at this temperature for 5 min.
The flow rate of the carrier gas helium was 2 mL/min. The ionization
energy in the electron impact mode was 70 eV.
Determination of Response Factors for LC-MS/MS. Solutions
of coumarin and [13C2]-coumarin in aqueous formic acid (0.1%) were
mixed in molar ratios ranging from 0.1 to 9 to give a total volume of
10 mL and a total coumarin content of 3 µg. Subsequently, the coumarin
mixtures were subjected to LC-MS/MS as outlined before. Response
factors Rf were calculated as reported recently (17) and gave a response
factor of 0.90 for the SRM transition 147/103 and 149/104 for unlabeled
and labeled coumarin, respectively.
Because the recently developed SIDAs of mycotoxins and
vitamins were based on LC-MS/MS as a result of matrix

Stable Isotope Dilution Assay of Coumarin
J. Agric. Food Chem., Vol. 56, No. 3, 2008 799
Figure 5. LC-MS/MS chromatograms of cinnamon star cookies in positive
electrospray ionization mode after CID of the protonated molecules. The
internal standard [13C2]-coumarin is detected in the trace MS/MS 149/
104 and unlabeled coumarin in trace MS/MS 147/103. UV: UV absorption.
Figure 4. Hypothetical fragmentation pathways of coumarin isotopologues
in electron impact ionization (A) and CID of the protonated molecules in
positive electrospray ionization mode (B).
interferences (14, 15), tandem MS was also applied to coumarin
isotopologues, being necessary for unequivocal quantification.
When CID was employed to [M + H]+ of isotopologic
coumarins, the spectra shown in Figure 2 were obtained.
Besides residual [M + H]+, two intense signals at m/z 103 and
91 and m/z 104 and 91 for coumarin and [13C2]-coumarin,
respectively, are discernible. The two signals for each coumarin
isotopologue can be assigned to the (a) loss of one CO2 retaining
one 13C label in [13C2]-coumarin and (b) loss of two CO
resulting in unlabeled benzylium (4) or tropylium (5) for both
labeled and unlabeled coumarin (Figure 4B). Because the
product ion of path (a) retains one label, the latter signal was
used for differentiation and quantification of the isotopologues.
The calibration by measuring different ratios of the isotopo-
logues revealed a constant response factor for the product ions
at m/z 103 and 104 of [M + H]+ in MS/MS over 2 decades of
isotope ratios.
Figure 6. LC-MS/MS chromatograms of Ceylon cinnamon.
Extraction and Analysis of Free and Total Coumarin by
LC-MS/MS. Because LC-tandem MS was specific enough
to differentiate coumarin isotopologues from sample interfer-
ences, sample preparation for LC-MS/MS of free coumarin
proved to be very straightforward. After the samples were
homogenized in 80% methanol containing definite amounts of
[13C2]-coumarin, the extracts only had to be filtered and passed
through a syringe filter. LC-MS/MS of cinnamon star cookies
(Figure 5), Ceylon cinnamon powder (Figure 6), Ceylon
cinnamon bark essential oil (Figure 7), and curry spice (Figure
8) displayed in ESI-MS conceivable signals for the isotopo-
logues in their respective selected reaction monitoring (SRM)
traces devoid from matrix peaks. In particular, when the
chromatograms of Ceylon cinnamon (Figure 6) and curry spice
(Figure 8) were regarded, the UV signal at 275 nm reveals a
myriad of signals, which hardly can be differentiated from the
small coumarin signal.
part of the sum (henceforth referred to as “total” coumarin) of
free and liberable coumarin is assumed to occur in its free form.
To prove this assumption, we examined two different methods
to analyze total coumarin. The chemical alternative included
treatment with hydrochloric acid at 90 °C, whereas the
enzymatic counterpart consisted of treatment with ꢀ-glucosidase
from almonds (emulsin) after extraction. When these methods
were applied to various foods containing coumarin, we obtained
the results depicted in Table 1. For both cassia and cinnamon,
we found no significant amounts of bound coumarin. However,
in cinnamon star cookies, only 90% of coumarin occurred in
its free form, thus indicating that the cassia used as an ingredient
contained small amounts of the bound precursor. In contrast to
this, for fenugreek seeds higher coumarin contents were found
after both hydrochloric acid and ꢀ-glucosidase treatment.
Obviously, about 68% were present as bound precursor, which
was hydrolyzed effectively both by acidic and enzymatic
hydrolysis. However, in curry spice, the coumarin content of
which originated from fenugreek seeds as ingredients, ꢀ-glu-
Generally, plants have been reported to liberate coumarin
from its precursor coumarinyl glucoside (20) by enzyme action
after disrupture of cells. For example, in woodruff, only a minor

800 J. Agric. Food Chem., Vol. 56, No. 3, 2008
Rychlik
Table 2. Total Content of Coumarin in Foods
range of
total coumarin
(mg/kg)
food (number of samples)
cassia cinnamon (Cinnamomum aromaticum), powdered (3)
Ceylon cinnamon (Cinnamomum ceylanicum), powdered (1)
Ceylon cinnamon (Cinnamomum ceylanicum), bark (1)
cassia (Cinnamomum aromaticum), bark essential oil (1)
cinnamon (Cinnamomum ceylanicum), bark essential oil (1)
cinnamon capsule (1)
1250–1490
2.4
0.86
4370
40
292
cinnamon star cookies (2)
19–65
7.8–23
7.3
pastries with cinnamon except cinnamon star cookies (2)
breakfast cereals with cinnamon (1)
rice pudding with cinnamon (1)
0.17
cinnamon tea (2)
mulled wine (1)
woodruff (Asperula odorata) (1)
0.736–0.94
0.036
203
blue-white clover seeds (Trigonella caerulea) (1)
fenugreek seeds (Trigonella foenicum graecum) (1)
curry spice (1)
cookies without cinnamon, commercial products (2)
green tea (Thea sinensis), powder (1)
soft drink with green tea (1)
37
0.34
0.46
nda-0.03
0.21
0.07
Figure 7. LC-MS/MS chromatograms of Ceylon cinnamon bark oil.
lavender (Lavandula angustifolia), essential oil (1)
chamomille (Matricaria recutita), essential oil (1)
citron (Citrus limon), essential oil (1)
coriander (Coriandrum sativum), essential oil (1)
dill (Anethum graveolens) seeds, essential oil (1)
peppermint (Mentha x piperita), essential oil (1)
bilberries (Vaccinium myrtillus) (1)
raspberries, black (Ribes nigrum) (1)
carrots (Daucus carota), raw (1)
Angelica roots (Angelica archangelica) (1)
124
3.23
3.86
0.30
0.21
0.30
nda
nda
nqa
nda
a nd, below limit of detection; nq, below limit of quantitation.
and especially in baked goods, the detection limit (DL) was
determined in butter cookies according to the method of Hädrich
and Vogelgesang (15). The calculations resulted in a DL of 2.9
µg/kg and a quantification limit of 8.6 µg/kg in starch containing
foods, which proved to be sufficient as the legal limit for
coumarin content in foods is as high as 2 mg/kg. In comparison
to HPLC-UV, the new method appeared to be 3 orders of
magnitude more sensitive. Recovery was evaluated by adding
10 µg/kg to butter cookies and was found to be 94.1%.
Intra-assay precision was determined by analyzing coumarin
in breakfast cereals and revealed a coefficient of variation of
9.9% for total coumarin (n ) 5).
Figure 8. LC-MS/MS chromatograms of curry spice after treatment with
glucosidase.
Table 1. Free and Total Content of Coumarin in Foods
Coumarin Contents in Foods. To prove the suitability of
the new method and to verify or disprove coumarin occurrence
in different foods, we applied it to a variety of products and
acquired the data listed in Table 2. The scope of this survey
was not to obtain a representative range of contents in the single
foods but to obtain an insight as to whether previous studies
might have reported inaccurate data because of insufficient
methodology.
coumarin (mg/kg)
without
after treatment
after glycosidase
treatment
sample
hydrolysis with hydrochloric acid
cassia
1380
2.5
61
0.11
0.29
24
1450
2.4
68
0.31
0.33
naa
1490
2.4
72
0.34
0.46
203
Ceylon cinnamon
cinnamon star cookies
fenugreek seeds
curry spice
woodruff
The highest total coumarin content exceeding 4 g/kg was
found in cassia bark oil, which is well in line with the high
abundance of the odorant reported in the literature (10).
However, our data are substantially lower than the published
high contents ranging between 16 and 25 g/kg. Moreover,
concentrations exceeding the g/kg range were detected in cassia,
which is in agreement with data obtained by He et al. (19). In
contrast to this, we corroborated the findings of Miller et al.
(20) and Ehlers et al. (10), who found far lower contents in
Ceylon cinnamon. In the powdered Ceylon cinnamon powder
and the bark essential oil, we quantified coumarin concentrations
of about 3 orders of magnitude below those of the respective
cassia products. In many foods that are labeled to contain
a na ) not analyzed.
cosidase was more effective in liberating the precursor. Contrary
to the seeds, the percentage of free coumarin in curry spice was
as high as 63%. The highest amount of 88% bound coumarin
was found in woodruff. Because liberation of coumarin was
easily initiated during homogenizing the tissue and instant
activity of endogenous ꢀ-glucosidase, the free content was only
quantifyable when inactivating the enzyme by CaCl2 treatment
during sample homogenization.
Performance Criteria. To evaluate whether sensitivity of
LC-MS/MS was sufficient for quantifying coumarin in foods

Stable Isotope Dilution Assay of Coumarin
J. Agric. Food Chem., Vol. 56, No. 3, 2008 801
(5) European Council. Directive 88/388/EEC, 1988, http://ec.europa.
eu/food/fs/sfp/addit_flavor/flav09_en.pdf.
(6) Khan, A.; Safdar, M.; Khan, M. M.; Khattak, K. N.; Anderson,
R. A. Cinnamon improves glucose and lipids of people with type
2 diabetes. Diabetes Care 2003, 26, 3218.
(7) AFC Opinion of the Scientific Panel on Food Additives, Flavour-
ings, Processing Aids and Materials in Contact with Food (AFC)
on a request from the Commission related to Coumarin. EFSA J.
2004, 104, 1–36.
(8) Walters, D. G.; Lake, B. G.; Cottrell, R. C. High-performance
liquid chromatography of coumarin and its metabolites. J. Chro-
matogr. 1980, 196, 501–505.
cinnamon, coumarin was detected in the mg/kg range. Therefore,
in these products, cassia must have been used to generate the
flavor of cinnamon. Particularly in cinnamon star cookies,
coumarin was highly abundant with concentrations amounting
to 65 mg/kg. In cinnamon-flavored tea and in mulled wine, the
coumarin concentration in the beverages was below 1 mg/kg,
indicating that only small amounts of cassia have been included
as an ingredient. One sample of rice pudding with cinnamon
contained only 0.17 mg/kg coumarin and proved that the
manufacturers have changed their recipe toward an omission
of cassia.
(9) Ehlers, D.; Pfister, M.; Bork, W.-R.; Toffel-Nadolny, P. HPLC
analysis of tonka bean extracts. Z. Lebensm.-Unters. Forsch. 1995,
201, 278–282.
(10) Ehlers, D.; Hilmer, S.; Bartholomae, S. HPLC analysis of
supercritical carbon dioxide cinnamon and cassia extracts in
comparison with cinnamon and cassia oils. Z. Lebensm.-Unters.
Forsch. 1995, 200, 282–288.
(11) Ranfft, K.; Gerstl, R. Gas chromatographic method for the
determination of coumarin in melilotus. Wirtschaftseigene Futter
1974, 20, 250–257.
(12) Christakopoulos, A.; Feldhusen, K.; Norin, H.; Palmqvist, A.;
Wahlberg, I. Determination of natural levels of coumarin in
different types of tobacco using a mass fragmentographic method.
J. Agric. Food Chem. 1992, 40, 1358–1361.
(13) Jessome, L. L.; Volmer, D. A. Ion suppression: A major concern
in mass spectrometry. LCGC North America 2006, 2006, 83–89.
(14) Rychlik, M. Pantothenic acid quantification by a stable isotope
dilution assay based on liquid chromatography-tandem mass
spectrometry. Analyst 2003, 128, 832–837.
(15) Asam, S.; Rychlik, M. Quantitation of type B-trichothecene
mycotoxins in foods and feeds by a multiple stable isotope dilution
assay. Eur. Food. Res. Technol. 2006, 224, 769–783.
(16) Perkin, W. Deutsch. Chem. Ges. 1875, 8, 1599.
(17) Rychlik, M. Quantification of free and bound pantothenic acid in
foods and blood plasma by a stable isotope dilution assay. J. Agric.
Food Chem. 2000, 48, 1175–1181.
(18) Hädrich, J.; Vogelgesang, J. Concept 2000—A statistical approach
for analytical practice—Part 1: Limits of detection, identification
and determination. Dtsch. Lebensm. Rundsch. 1999, 95, 428–436.
(19) Porter, Q. N. Mass Spectrometry of Heterocyclic Compounds;
Wiley: New York, 1985.
(20) Edwards, K. G.; Stoker, J. R. Biosynthesis of coumarin: The
isomerization stage. Phytochemistry 1967, 6, 655–661.
(21) Von Sydow, E.; Anjou, K. Aroma of bilberries (Vaccinium
myrtillus). I. Identification of volatile compounds. Lebensm. Wiss.
Technol. 1969, 2, 78–81.
Further foods containing substantial amounts of coumarin
were woodruff, lavender essential oil, and blue-white clover
seeds. In contrast to this, the levels in fenugreek seeds as the
ingredient of many spices was smaller than 1 mg/kg and, in
consequence, curry spice was also low in coumarin. The
occurrence of the odorant in chamomille and citron essential
oil was confirmed, whereas in green tea powder, carrots, and
the essential oils of coriander, dill, and peppermint coumarin
was present only in minute amounts. Moreover, earlier papers
on coumarin occurring in bilberries, raspberries, carrots, and
Angelica roots were not corroborated (21–23).
Presumably, coumarin in Angelica was mixed up with
coumarin derivatives, such as umbelliferon and its glucosides.
In flavor research, many successful applications of SIDAs
based on GC-MS detection have been reported. However, gas
chromatography requires separation of the odorants from
nonvolatile compounds, which demands at least one further
distillation step. The present study confirmed the merits of SIDA
methodology in respect to specifity, sensitivity, and accuracy
also in an LC-MS instrumentation. Once more, the simple and
fast sample preparation points to the superiority of the present
assay, which is the first paper on accurate coumarin quantitation
in trace amounts and in complex matrices, such as baked
products.
ACKNOWLEDGMENT
I am grateful to Mrs. D. Fottner for her excellent technical
assistance. Appreciation is also due to Mrs. I. Otte and Mr. S.
Kaviani for running the LC-MS/MS equipment.
LITERATURE CITED
(1) Cox, D.; O’Kennedy, R.; Thornes, R. D. The rarity of liver toxicity
in patients treated with coumarin (1,2-benzopyrone). Hum. Toxicol.
1989, 8, 501–506.
(22) Coppens, A.; Hoejenbos, L. Volatile constituents of raspberry juice
(Rubus idaeus L.). Rec. TraV. Chim. Pays-Bas Belg. 1939, 58,
675–679.
(23) Seifert, R. M.; Buttery, R. G.; Ling, L. Identification of some
constituents of carrot seed oil. J. Sci. Food Agric. 1968, 19, 383–
385.
(2) Shilling, W. H.; Crampton, R. F.; Longland, R. C. Metabolism
of coumarin in man. Nature 1969, 221, 664–665.
(3) Born, S. L.; Caudill, D.; Fliter, K. L.; Purdon, M. P. Identification
of the cytochromes P450 that catalyze coumarin 3,4-epoxidation
and 3-hydroxylation. Drug Metab. Dispos. 2002, 30, 483–487.
(4) Born, S. L.; Hu, J. K.; Lehman-Mckeeman, L. D. o-Hydroxyphe-
nylacetaldehyde is a hepatotoxic metabolite of coumarin. Drug
Metab. Dispos. 2000, 28, 218–223.
Received for review September 24, 2007. Revised manuscript received
December 3, 2007. Accepted December 5, 2007.
JF0728348